Penn ESE534 Spring2014 -- Mehta & DeHon 1 ESE534 Computer Organization Day 6: February 12, 2014 Energy, Power, Reliability

Penn ESE534 Spring2014 -- Mehta & DeHon1

ESE534Computer Organization

Day 6: February 12, 2014

Energy, Power, Reliability


Today

• Energy tradeoffs

• Voltage limits and leakage

• Variations

• [more of the gory transistor equations…]


At Issue

• Many now argue energy will be the ultimate scaling limit– (not lithography, costs, …)

• Proliferation of portable and handheld devices– …battery size and life biggest issues

• Cooling, energy costs may dominate cost of electronics – Even server room applications

Preclass 1

• 1GHz case– Voltage?– Energy per Operation?– Power required for 2 processors?

• 2GHz case– Voltage?– Energy per Operation?– Power required for 1 processor?


Preclass 1 Lesson

• What does Preclass 1 result tell us?



Energy and Delay

2

2

1CVE

gd=Q/I=(CV)/I

Id,sat=(COX/2)(W/L)(Vgs-VTH)2


Energy/Delay Tradeoff

• EV2

• gd1/V

• We can trade speed for energy

• E×(gd)2 constantMartin et al. Power-Aware Computing, Kluwer 2001 http://caltechcstr.library.caltech.edu/308/

2

2

1CVE

gd=(CV)/IId,sat (Vgs-VTH)2


Area/Time Tradeoff

• Also have Area-Time tradeoffs– HW2 spatial vs temporal multipliers– See more next week

• Compensate slowdown with additional parallelism

• …trade Area for Energy Architectural Option


Question

• By how much can we reduce energy?

• What limits us?


Challenge: Power


Origin of Power Challenge

• Limited capacity to remove heat– ~100W/cm2 force air– 1-10W/cm2 ambient

• Transistors per chip grow at Moore’s Law rate = (1/F)2

• Energy/transistor must decrease at this rate to keep constant power density

• P/tr CV2f• E/tr CV2

– …but V scaling more slowly than F


ITRS Vdd Scaling:More slowly than F


ITRS CV2 Scaling:More slowly than (1/F)2


Origin of Power Challenge

• Transistors per chip grow at Moore’s Law rate = (1/F)2

• Energy/transistor must decrease at this rate to keep constant

• E/tr CV2


Historical Power Scaling

[Horowitz et al. / IEDM 2005]

Microprocessor Power Density


The Future of Computing Performance: Game Over or Next Level?National Academy Press, 2011

http://www.nap.edu/catalog.php?record_id=12980

Watts

Intel Power Density



Source: Carter/Intel

18

Power Limits Integration

Impact

Impact

• Power density is limiting scaling– Can already place more transistors on a chip than

we can afford to turn on!

• Power is potential challenge/limiter for all future chips.– Only turn on small percentage of transistors?– Operate those transistors as much slower

frequency?

– Find a way to drop Vdd?



How far can we reduce Vdd?


Limits

• Ability to turn off the transistor

• Parameter Variations

• Noise (not covered today)

MOSFET Conduction

Penn ESE534 Spring2014 -- Mehta & DeHon22From: http://en.wikipedia.org/wiki/File:IvsV_mosfet.png

Transistor Conduction


• Three regions– Subthreshold (Vgs<VTH)

– Linear (Vgs>VTH) and (Vds < (Vgs-VTH))

– Saturation (Vgs>VTH) and (Vds > (Vgs-VTH))

Saturation Region


Ids,sat=(COX/2)(W/L)(Vgs-VTH)2

• (Vgs>VTH)

• (Vds > (Vgs-VTH))

Linear Region


Ids,lin=(COX)(W/L)((Vgs-VTH)Vds-(Vds)2/2)

• (Vgs>VTH)

• (Vds < (Vgs-VTH))


Subthreshold Region

• (Vgs<VTH)

Isub IVT 10Vgs VTH / S

qkTS /))10(ln( [Frank, IBM J. R&D v46n2/3p235]

Operating a Transistor

• Concerned about Ion and Ioff

• Ion drive (saturation) current for charging

– Determines speed: Tgd = CV/I

• Ioff leakage current

– Determines leakage power/energy:

• Pleak = V×Ileak

• Eleak = V×Ileak×Tcycle



Leakage

• To avoid leakage want Ioff very small

• Switch V from Vdd to 0

• Vgs in off state is 0 (Vgs<VTH)


Ioff IVT 10 VTH / S


Leakage

• S90mV for single gate• S70mV for double gate

• For lowest leakage, want S small, VTH large

• 4 orders of magnitude IVT/IoffVTH>280mV

Ioff IVT 10 VTH / S

Leakage limits VTH in turn limits Vdd

How maximize Ion/Ioff ?

• Maximize Ion/Ioff – for given Vdd ? EswCV2

• Get to pick VTH, Vdd



Id,lin=(COX)(W/L)(Vgs-VTH)Vds-(Vds)2/2

Id,sat=(COX/2)(W/L)(Vgs-VTH)2

Preclass 2

• E = Esw + Eleak

• Eleak = V×Ileak×Tcycle

• EswCV2

• Ichip-leak = Ndevices ×Itr-leak



Preclass 2

• Eleak(V) ?

• Tcycle(V)?


In Class

• Assign calculations– SIMD – each student computes for a

different Voltage

• Collect results on board

• Group: Identify minimum energy point and discuss


Graph for In Class


Impact

• Subthreshold slope prevents us from scaling voltage down arbitrarily.

• Induces a minimum operating energy.



Challenge: Variation


Statistical Dopant

Count and Placement

[Bernstein et al, IBM JRD 2006]


Vth Variability @ 65nm

[Bernstein et al, IBM JRD 2006]

Penn ESE534 Spring2014 -- Mehta & DeHon 40

Variation• Fewer dopants, atoms increasing Variation• How do we deal with variation?

40

% variation in VTH

(From ITRS prediction)


Impact of Variation?

• Higher VTH?

– Not drive as strongly slower

– Id,sat (Vgs-VTH)2

• Lower VTH?

– Not turn off as well leaks more

Ioff IVT 10 VTH / S


Variation• Margin for expected variation• Must assume VTH can be any value in range

Pro

babi

lity

Dis

trib

utio

n

VTH

Ion,min=Ion(Vth,max)Id,sat (Vgs-VTH)2

Vgs = Vdd


Margining

• Must raise Vdd to increase drive strength

• Increase energy

Pro

babi

lity

Dis

trib

utio

n

VTH

Ion,min=Ion(Vth,max)Id,sat (Vgs-VTH)2

Vgs = Vdd

Penn ESE534 Spring2014 -- Mehta & DeHon 44

Variation

• Increasing variation forces higher voltages– On top of our leakage limits

44


• Margins growing due to increasing variation

• Margined value may be worse than older technology?

Variations

Pro

babi

lity

Dis

trib

utio

n

Delay

New

Old


End of Energy Scaling?

[Bol et al., IEEE TR VLSI Sys 17(10):1508—1519]

Black nominalGrey with variation

Chips Growing

• Larger chips (billions of transistors) sample further out on distribution curve


From: http://en.wikipedia.org/wiki/File:Standard_deviation_diagram.svg


Big Ideas

• Can trade time for energy– … area for energy

• Variation and leakage limit voltage scaling• Power major limiter going forward

– Can put more transistors on a chip than can switch

• Continued scaling demands – Deal with noisier components

• High variation• … other noise sources

Admin

• Homework 2 due Tonight

• Reading for Monday on web

• Homework 3 due next Wednesday


Documents

Penn ESE534 Spring2014 -- Mehta & DeHon 1 ESE534 Computer Organization Day 6: February 12, 2014 Energy, Power, Reliability